Combined experimental and numerical approach for identification of dynamic material model parameters

Extraction of the material stress-strain curve from a dynamic tensile or shear experiment is not straightforward. Indeed, stress and strain are not homogeneously distributed in the specimen, and consequently no one-one relation exists between the measured elongation and strain on one hand, and the measured force and stress on the other hand. This work aims at improving the accuracy of the stress-strain curves calculated from high strain rate experiments and the modelling of the material behaviour. Therefore numerical simulations are used to determine the relationship between the average stress-strain and local effective stress-strain. The material model parameters used in these simulations are improved during an iterative procedure which combines the experimental results and the simulated stress and strain distribution. Stress triaxiality, local temperature and strain rate are taken into account. The method is applied to dynamic tensile and shear experiments on a Ti6Al4V alloy carried out on a split Hopkinson bar set up. The Johnson-Cook model is used to describe the strain rate and temperature dependent material behaviour. The two types of tests are used separately or simultaneously to extract and model the material behaviour. It is found that using tensile and shear experiments simultaneously has clear advantages. The same approach is used to identify parameters for the Johnson-Cook damage initiation criteria.